Hi Priya, If you want a stablized 'checkpoint size', you should enable the state expiration by configuring the table.exec.state.ttl option to a value large than 0.
See more in: https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/config/#table-exec-state-ttl If you want a stablized 'incremental checkpoint size', this might be difficult to make. When performing incremental checkpoint, RocksDBStateBackend will upload the 'new created SST files' to remote storage. The total size of these files is the incremental checkpoint size. However, the new created SST files generated by RocksDB's compaction are affected by the compaction strategy and will not have a fixed size. Of course, if you disable incremental checkpoint, the 'incremental checkpoint size' will equal to 'checkpoint size'. Hope this helps you. Xiangyu Feng banu priya <banuke...@gmail.com> 于2024年8月5日周一 15:58写道： > Hi All, > > I have a flink job with Rabbitmq source, sink,keyby, tumbling window and > aggregator function. Incremental rocksdb checkpointing is enabled. > I understand that checkpoints size will become stabilized at some point > and it won't increase after that. > > How can I prove that to my customer?. > > I ran my job for 7 days it increased to 250MB. Not sure even if I run for > 30 days it will decrease. > > I saw in one of AWS documentation that if we disable incremental > checkpoints we may find the size at which it will stabilize quickly. Is > that true?? > > Thanks > Banu >
